Atom counting with accelerator mass spectrometry

Accelerator mass spectrometry (AMS) was born in the late 1970s, when it was realized at nuclear physics laboratories that the accelerator systems can be used as a sensitive mass spectrometer to measure ultralow traces of long-lived radioisotopes. It soon became possible to measure radioisotope-to-stable-isotope ratios in the range from ${10}^{-12}$ to ${10}^{-16}$ by counting the radioisotope ions “atom by atom” and comparing the count rate with ion currents of stable isotopes ($1.6\mathrm{\mu }\mathrm{A}=1\times {10}^{13}$ singly charged ions/s). It turned out that electrostatic tandem accelerators are best suited for this, and there are now worldwide about 160 AMS facilities based on this principle. This review presents the history, technological developments, and research areas of AMS through the 45 yr since its discovery. Many different fields are touched by AMS measurements, including archaeology, astrophysics, atmospheric science, biology, climatology, cosmic-ray physics, environmental physics, forensic science, glaciology, geophormology, hydrology, ice core research, meteoritics, nuclear physics, oceanography, and particle physics. Since it is virtually impossible to discuss all fields in detail in this review, only specific fields with recent advances are highlighted in detail. For the others, an effort is made to provide relevant references for in-depth studies of the respective fields.

Figure 1

The 60 in. cyclotron at the Lawrence Radiation Laboratory, Berkeley, soon after completion in 1939. The key figures in its development and use are, from left to right, Dr. D. Cooksey, Dr. D. Corson, Dr. Ernest Orlando Lawrence (the inventor of the cyclotron), Dr. R. Thornton, Dr. J. Backus, W. S. Sainsbury, Dr. L. W. Alvarez, and Dr. Edwin Mattison McMillan.

Figure 2

Production rate of long-lived radionuclides by cosmic rays in the atmosphere. The main target atoms for the production of the respective radionuclides are indicated with different colors. The atmospheric concentration of ${\mathrm{CO}}_2$ in 2022 is 0.042%. The total production rates in the atmosphere ($\mathrm{atoms}/\mathrm{s}$) can be obtained by multiplying the plotted numbers with the surface of Earth ($5.1\times {10}^{18}{\mathrm{cm}}^2$). Adapted from [271].

Figure 3

Schematic presentation of ${}^{14}\mathrm{C}$ production in the atmosphere through the nuclear reaction ${}^{14}\mathrm{N}(\mathit{n},\mathit{p}){}^{14}\mathrm{C}$, with neutrons originating from both cosmic-ray-induced spallation of atmospheric nuclei and man-made nuclear explosions. The subsequent two-step oxidation of ${}^{14}\mathrm{C}$ to ${{}^{14}\mathrm{CO}}_2$ and its uptake by the biosphere and the ocean are also indicated. The photo is from the first hydrogen-bomb test “Ivy Mike” in 1952 (National Nuclear Security Administration, Nevada Site Office). Adapted from [563].

Figure 4

Schematic layout of the VERA AMS facility in the configuration of 2022. The original facility became operational in 1997 and has since gone through several upgrades in order to accommodate the presently detected isotopes. From [186].

Figure 5

Display of radionuclides that are of interest for AMS measurements. Radionuclides that can be measured at smaller tandem AMS facilities like VERA are marked in green, whereas those that can currently be measured only at larger tandem facilities are marked in brown. The ones marked in purple require positive-ion production and acceleration via heavy-ion linear accelerators ([95]; [248]; [504]) or heavy-ion cyclotrons ([94]). The ones marked in red can now be measured at VERA using the new laser-anion interaction system ILIAMS ([335]; [336], [338]); see also Sec. 2d3. Previously ${}^{90}\mathrm{Sr}$ could be measured only at a large tandem facility ([396]). For radionuclides marked in blue, a stable isobar suppression has presently not been achieved.

Figure 6

Illustration of the size reduction of AMS facilities strongly depending on the terminal voltage of the tandem accelerator. Terminal voltage and floor space requirement are indicated with red labels. Note that “Mini AMS” is now known as the mini radiocarbon dating system (MICADAS). Adapted from [272].

Figure 7

(a) Schematic illustration of a custom-built Middleton-type high-intensity negative-ion source used by the Hebrew University AMS group ([178]). Cs vapor delivered into the source volume by an oven (internal in this design) and ${\mathrm{Cs}}^{+}$ ions produced on the resistively heated hemispherical ionizer surface (tantalum) are accelerated and focused onto the target at a negative potential of about 5 kV. The solid AMS sample is pressed in a 1.5 or 1 mm diameter hollow in the conical target, which is cooled by Freon circulation. Negative ions ejected by the same potential through the central 4 mm diameter hole in the ionizer are eventually extracted and transported toward the injection system of the accelerator. (b) Photographs (from left to right) of the ion source components: (i) A sample wheel containing positions for 28 targets. The target is rotated using a computer-controlled pneumatic motor for insertion into the source body. (ii) Ion source assembly. The wheel is located in the flat box (separately pumped via the lower port) and can be isolated from the ion source volume by a hand gate valve (black handle) for fast replacement. (iii) Ion source chamber with its extractor insulator (white).

Figure 8

Example of a compact AMS system. The 4103Bo AMS multielement system is shown operating with a vacuum insulated tandem accelerator as produced by HVE. The vacuum insulated tandem accelerator principle was pioneered at ETH Zurich ([498]) and is being used in the MICADAS facilities ([499]); it is now manifuctured by Ionplus. In our example, a 200-sample sputter ion source delivers negative ions that are energy (electrostatic analyzer) and mass filtered (90° magnet) and injected into the tandem accelerator. Collisions in the gas stripper at the terminal strip off electrons and eliminate molecular ions. Positively charged ions are then accelerated a second time, and the mass and energy are again analyzed at the high-energy side of the system. A degrader foil (such as for Be AMS) results in isobars with different energies (higher energy loss for B than for Be) that are then separated by the electrostatic deflector. A second 120° magnet reduces the background further, including isotopic interference, and often does not require an additional TOF system (even for actinide measurements). The rare radionuclides are counted in the final gas-ionization chamber, while stable isotopes are measured as currents after the low-energy and high-energy magnets. Adapted from [462].

Figure 9

Simulation of isobar reduction for ${}^{10}\mathrm{Be}$ AMS via a thick absorber in front of the detector. A stack of silicon nitride membranes stops boron with its higher atomic number ($Z=5$), while beryllium ($Z=4$) enters the ionization chamber, where they are stopped in the gas volume. The remaining background (scattered ions or residual ${}^{10}\mathrm{B}$) can be identified by the differing energy losses along the two anodes of the chamber. Adapted from [482].

Figure 10

Left panel: isobar reduction of ${}^{26}\mathrm{Mg}$ via laser-ion interaction and a gas reaction cell for ${}^{26}\mathrm{AlO}$ AMS. The molecular isobar ${\mathrm{MgO}}^{-}$ is suppressed by $\sim 5$ orders of magnitude in the simple passage through a He-filled ion cooler. Adding an appropriate laser for selective photodetachment of ${\mathrm{MgO}}^{-}$ results in additional suppression of up to 11 orders of magnitude and combined yields a suppression factor exceeding ${10}^{14}$ at an injected laser power of 12 W. The use of ${\mathrm{AlO}}^{-}$ increases the detection efficiency relative to the use of ${\mathrm{Al}}^{-}$ extraction by a factor of between 3 and 5. Adapted from [290]. Right panel: ${}^{36}\mathrm{Cl}$ AMS. Effect of the laser interaction for suppression of ${{}^{36}\mathrm{S}}^{-}$: both ${}^{36}\mathrm{Cl}$ and ${}^{36}\mathrm{S}$ reach the detector without a laser (gray circles). If the laser is turned on (blue symbols), only ${}^{36}\mathrm{Cl}$ events are recorded (here at a laser power of 1 W). Adapted from [289].

Figure 11

Schematic illustration of the gas-filled magnet principle. Trajectories followed by heavy ions in a magnetic field region are shown. Top panel: trajectory in vacuum. Two isobaric ions (blue and red) with the same charge states have degenerate trajectories [unless under exceptionally high magnetic resolving power $(m/q)/d(m/q)\sim {10}^5$]. Bottom panel: trajectory in a gas-filled magnetic field region. The ions follow two broadened paths governed by their mean ion charge state in the gas that are different owing to their differing ionic charges as a result of differences in $Z$, and additionally owing to the different energy loss that they experience through their flight through the gas.

Figure 12

(a) Depth profile of the differential energy loss in a multianode ionization chamber for ${}^{41}\mathrm{Ca}$, its isobar ${}^{41}\mathrm{K}$, and a neighboring isotope ${}^{42}\mathrm{Ca}$. The profile is probed by the successive $\mathrm{\Delta }E$ and $E_{\mathrm{RES}}$ anodes. Adapted from [157].

Figure 13

(a) Schematic illustration (side and top views) of the multianode ionization chamber used at Australian National University for AMS detection of ${}^{53}\mathrm{Mn}$ and ${}^{60}\mathrm{Fe}$. (b) Photograph of the detector showing (from bottom to top) anode, grid, and cathode planes. Dimensions are in mm. Adapted from [334].

Figure 14

(a) Schematic drawing of a compact ion chamber (dimensions in mm). (b) The assembly as an insertable-retractable version, as applied at ANU, Canberra ([334]), and based on designs developed at ETH Zurich ([492]; [367]) and VERA ([165]). (c) Diagram of a compact gas counter for AMS at low energies [provided by ETH and installed at CNA (Spain)] mounted at the end of a beamline. Low-noise preamplifiers are mounted directly on the electrical feedthrough. The detector housing is electrically insulated. Adapted from [461].

Figure 15

Schematic diagram of the electron cyclotron resonance ion source originally developed by [179], [180]; see the text. The ellipse shown in the plasma chamber depicts the cross section of the surface onto which the electron cyclotron resonance [Eq. (1)] is met. Adapted from [399].

Figure 16

Schematic illustration of the ATLAS facility and the Enge gas-filled magnetic spectrograph for ${}^{39}\mathrm{Ar}$ ($t_{1/2}=269\mathrm{yr}$) detection by AMS ([504]; [399]). The ECR sources ECR II and ECR III, presently in use, are shown; ECRCB is an ECR charge breeder for externally fed ions. The rare ions (${}^{39}\mathrm{Ar}$) are identified and counted in a position-sensitive detector and ionization chamber along the spectrograph focal plane; see Fig. 17. A retractable Faraday cup at the target position (FC-T) is located in front of the spectrograph and is used to measure the charge current of a stable Ar ion (${}^{38}\mathrm{Ar}$ or ${}^{40}\mathrm{Ar}$), allowing one to measure the transmission efficiency between a front end Faraday cup (FC-I) and target position for stable isotopes and, eventually, the quantitative determination of the isotopic ratio ${}^{39}\mathrm{Ar}/\mathrm{Ar}$. In addition, a rotatable Si detector (Si) is used to optimize the transmission of the rare ions.

Figure 17

Identification spectra of ions for an Ar sample extracted from deep oceanic water where the ${}^{39}\mathrm{Ar}$ concentration ${}^{39}\mathrm{Ar}/{}^{40}\mathrm{Ar}=2.6\times {10}^{-16}$ compared to atmospheric $8.1\times {10}^{-16}$ is a measure of the age of the water sample ($\sim 400\mathrm{yr}$) since atmospheric Ar dissolution. The ${}^{39}\mathrm{K}$ contaminant, spatially separated from ${}^{39}\mathrm{Ar}$ in the focal plane of the gas-filled magnet (see Fig. 11), was partially shielded in front of the focal-plane detector due to its extremely high count rate. Adapted from [95].

Figure 18

${}^{37}\mathrm{Ar}$ ions detected background free by a full stripping to $q={18}^{+}$ after midacceleration at the ATLAS positive-ion AMS.

Figure 19

Isobaric separation of ${}^{146}\mathrm{Sm}$ and stable ${}^{146}\mathrm{Nd}$ achieved with highly charged positive-ion AMS at the ATLAS facility (ANL) and the gas-filled Enge magnetic spectrograph. The Sm and Nd ions were accelerated to $6\mathrm{MeV}/u$. Adapted from [248].

Figure 20

(a) Schematic of the ATTA atomic beamline. Atoms (dotted line) travel from left to right, starting at the radio-frequency discharge source, passing through the laser-induced collimation and Zeeman slowing stages, and ending in the magneto-optical trap (MOT), where they are imaged using a CCD camera. (b),(c) Typical spectra of the trap capture rate of ${}^{81}\mathrm{Kr}$, ${}^{83}\mathrm{Kr}$, and ${}^{85}\mathrm{Kr}$ vs the laser frequency shift (relative to a reference transition in stable ${}^{84}\mathrm{Kr}$) measured for an atmospheric Kr sample (${}^{81}\mathrm{Kr}/\mathrm{Kr}=9.3\times {10}^{-13}$) ([569]). Adapted from [223].

Figure 21

Fluorescence signal from a single unambiguously identified atom of a noble-gas ${}^{81}\mathrm{Kr}$ atom in a MOT. Adapted from [316].

Figure 22

The growth of the number of tandem AMS facilities since 1978, prepared by [496] and formulated as a continuation of an earlier version ([495]). Prior to this, [154] presented a similar compilation in his talk during AMS11 in Rome. The colors of the column sections indicate the range of tandem terminal voltages in MV. Cyclotrons (green sections at the bottom) were in use only until 2008. The increase in the number of small tandem AMS facilities in the later years is clearly visible. At the top of the last column, the new development of positive-ion mass spectrometry (PIMS) facilities ([167]) is indicated. A more recent report on PIMS by the AMS group of SUERC is expected to appear in the Proceedings of the AMS-15 Conference in Sydney ([168]). The unique use of an ECR ion source coupled to a heavy-ion linear accelerator ([92]; [95]; [399]) is not depicted but is described in Sec. 2f.

Figure 23

World map with the global distribution of AMS machines. The size of the symbols indicates the number of machines in one particular location. The large symbols are transparent so as to allow smaller symbols to be visible if they overlap with the larger ones. For details on the various AMS facilities see Table 2. Courtesy of Christof Vockenhuber of ETH Zuerich.

Figure 24

The partly freed body of the Iceman as watched by two well-known mountain climbers from South Tyrol, Hans Kammerlander (left) and Reinhold Messner (right), on September 21, 1993. Kammerlander holds part of a wooden structure later identified as a carrying support of Oetzi. In the upper right corner the bow can be seen, its lower part stuck in the ice and its upper end leaning against the rocks. Just below the tip of the ski pole held by Messner one can see the smashed remains of a container made of bark from a birch tree, probably used to carry equipment for making a fire. Details of the equipment and their age determination with ${}^{14}\mathrm{C}$ AMS measurements were given by [285]. The photograph is courtesy of K. Fritz. From [281].

Figure 25

Details regarding the well-preserved rock art paintings of the Chauvet Cave in France, which were produced by prehistoric artists some 30 000 yr ago ([419]).

Figure 26

(a) Deviation of the atmospheric ${}^{14}\mathrm{C}$ content ($\mathrm{\Delta }{}^{14}\mathrm{C}$ in per mill) relative to a constant reference value ($\mathrm{\Delta }{}^{14}\mathrm{C}=0$) and displays $\mathrm{\Delta }{}^{14}\mathrm{C}$ for the last 4000 yr. (b) Annual tree rings from a local pine tree. The ${}^{14}\mathrm{C}$ content verified the bomb peak. Adapted from [475]. (c) Depiction of the sudden increase between 1955 and 1963 due to atmospheric nuclear weapons testing (cf. Fig. 3) and the gradual decrease after the Nuclear Test Ban Treaty of 1963. The latter is determined by the distribution of ${}^{14}\mathrm{C}$ due to the ${\mathrm{CO}}_2$ cycle. The slightly negative $\mathrm{\Delta }{}^{14}\mathrm{C}$ values before the rise of the bomb peak are due to the dilution of the natural ${{}^{14}\mathrm{CO}}_2$ content with ${}^{14}\mathrm{C}$-free ${\mathrm{CO}}_2$ from fossil fuel burning, which is known as the Suess effect ([488]).

Figure 27

Visualization of how the birth date of different cells in an individual of a known life span can be determined retrospectively by comparing the measured ${}^{14}\mathrm{C}$ signal from DNA extracts with the ${}^{14}\mathrm{C}$ bomb peak calibration curve. Based on work by [475]. Adapted from [276].

Figure 28

Evidence for neurogenesis in the human hippocampus. (a) $\mathrm{\Delta }{}^{14}\mathrm{C}$ values of (b) neuronal and (c) non-neuronal cells from the hippocampus plotted for the birth dates of 57 individuals. Deviations from the ${}^{14}\mathrm{C}$ bomb curve indicate the formation of new cells after birth. The offset between the atmospheric bomb curve and DNA measurements is indicative of the magnitude of cell turnover. The smaller offset of neurons indicates that their turnover is smaller than that of non-neurons. Based on work by [476]. Adapted from [276].

Figure 29

Two-year record of ${}^{14}\mathrm{CO}$ and CO concentration measurements in air from the observatory of Mt. Sonnblick (3106 m a.s.l.). The lower values during the summer are due to the depletion by reactions with increased OH concentrations. The unusually low values of both ${}^{14}\mathrm{CO}$ and CO (marked A and B) during winter are due to the transport of air from low latitudes to Mt. Sonnblick, as verified by four-day backward trajectory calculations ([439]). The unusually high CO value (marked C) is due to a contamination of the air by fossil fuel combustion from an emergency diesel engine running that day. Since this did not contribute any ${}^{14}\mathrm{C}$, the corresponding ${{}^{14}\mathrm{CO}}_2$ value of the contaminated air remains unaffected. From [439].

Figure 30

Model prediction for the development of the ${}^{14}\mathrm{C}$ content in the atmosphere from 1940 to 2100 expressed as the deviation of ${{}^{14}\mathrm{CO}}_2$ from the natural ${}^{14}{\mathrm{CO}}_2$ level ($\mathrm{\Delta }{}^{14}{\mathrm{CO}}_2=0$). The picture indicates the depression of ${}^{14}\mathrm{C}$ as a function of the so-called representative concentration pathways (RCPs). The numbers following RCP in the legend depict the radiation forcing in $\mathrm{W}/{\mathrm{m}}^2$, depending on different scenarios to reduce the emission of greenhouse gases ([519]). Adapted from [193].

Figure 31

Rapid ${}^{14}\mathrm{C}$ variations observed in tree rings occurring in 774 CE and 993 CE. Adapted from [356].

Figure 32

Great ocean conveyor as conceived originally by [51]. It is a simplified picture of the main ocean currents transporting heat around the globe. The warm and shallow currents are indicated in red, and the cold an salty deep current are marked in blue. Illustration by Joe Le Monnier, Natural History Magazine. Adapted from [271].

Figure 33

Distribution of almost 4000 profiling floats across the world oceans within the Argo Project. The colors of the dots indicate the country that released the floats. Continents are marked in bright green. Adapted from the [17].

Figure 34

Comparison of EPICA Dome C data with other paleoclimatic records ([135]). (a) Record of the insolation at high northern and southern latitudes ([32]). (b) Temperature record as measured with the $\delta D$ signal in the EPICA Dome C (blue curve) and the Vostok ice cores (red curve). The numbers refer to the marine isotope stage (MIS), where odd numbers were originally assigned to warm periods and even ones were assigned to cold periods ([134]; [467]). However, decimal numbers later became common to distinguish details of various subperiods ([227]). (c) Marine $\delta {}^{18}\mathrm{O}$ signals from two different deep-sea sediment cores (solid blue line and red dashed line) that reflect both temperature and ice mass changes. (d) The dust signal, which is particularly large during cold periods. Adapted from [271].

Figure 35

Chronology of ${}^{32}\mathrm{Si}$ half-life measurements. The solid line represents a mean value of 144 yr, with the calculations not including the ice and sediment values. The dashed line indicates a $1\sigma$ uncertainty of $\pm 11\mathrm{yr}$. Adapted from [148].

Figure 36

The periodic table of elements as established recently following the confirmation of discovery of elements up to the noble-gas oganesson ($Z=118$) and of their names and symbols adopted by the International Union of Pure and Applied Chemistry. The bold red frames denote elements with no stable isotopes and their longest lived isotope (in parentheses). Adapted from [312].

Figure 37

(a) AMS setup used at the Maier-Leibnitz Laboratory [Technical University of Munich (TUM)] for the search for superheavy elements in natural materials. (b) Identification spectrum of ions for an accelerator setting corresponding to hypothetical ${{}^{310}X}^{11+}$ ions. The rectangular contour shows the expected location of events, based on calibration of the residual energy measured by the Si detector and the ion time of flight through the detector system shown in (a). Adapted from [319].

Figure 38

Schematic curve of atomic abundances relative to that of Si as a function of atomic weight based on the data of [489], as presented by [65] in the B2FH review. The curve and the B2FH review where the major nucleosynthesis processes are identified had a major impact on the development of nuclear astrophysics. Processes denoted by $\mathrm{H}-$, He burning, and $\alpha$ refer mainly to exothermic fusion reactions up to the iron group; $s$ and $r$ refer, respectively, to the slow process occurring at low neutron densities in steady-state stellar conditions and rapid neutron-capture processes where extremely high neutron densities are produced in explosive stellar events. $P$ refers to the rare $p$-process stable nuclides (35 in number) that are produced via nuclear reactions other than neutron capture, such as photonuclear and charged-particle reactions. Adapted from [65].

Figure 39

AMS detection of ${}^{63}\mathrm{Ni}$ produced by the neutron-capture reaction ${}^{62}\mathrm{Ni}(n,\gamma ){}^{63}\mathrm{Ni}$. The neutron activation of an isotopically enriched ${}^{62}\mathrm{Ni}$ target was performed at Forschungs-Zentrum Karlsruhe with a quasi-Maxwellian neutron energy distribution ($kT=25\mathrm{keV}$) to study the $s$-process evolution in massive stars. The ${}^{63}\mathrm{Ni}$ ions, after separation of most of the stable isobaric ${}^{63}\mathrm{Cu}$ in a gas-filled spectrograph, are identified in the figure by the energy losses $\mathrm{\Delta }{E}_2$ and $\mathrm{\Delta }{E}_3$ measured on two successive anodes of a focal-plane ionization chamber. Adapted from [372].

Figure 40

Schematic representation of the $s$-process evolution starting from a ${}^{56}\mathrm{Fe}$ seed preexisting in a star. The process of successive neutron captures synthesizes about half of the heavy elements in nature. Inset: reproduced distribution of natural elements shown in Fig. 38 sorted by mass number. From [239].

Figure 41

Detection of ${}^{26}\mathrm{Al}$ produced by the charged-particle reaction ${}^{26}\mathrm{M}\mathrm{g}(p,n){}^{26\mathrm{g}}\mathrm{Al}$ and leading to the long-lived ground state ${}^{26\mathrm{g}}\mathrm{Al}$ ($t_{1/2}=7.2\times {10}^5\mathrm{yr}$). The reaction was studied at Argonne National Laboratory in the early years of AMS as the inverse of the destruction reaction of the important ${}^{26}\mathrm{Al}$ radionuclide. Adapted from [398].

Figure 42

Summary of ${}^{60}\mathrm{Fe}$ measurements in different archives. Adapted from [545].

Figure 43

Log-log plot of ${}^{41}\mathrm{Ca}$ vs ${}^{36}\mathrm{Cl}$ concentration measured in Antarctic meteorites of different terrestrial age. From the slope indicated and the known half-life of ${}^{36}\mathrm{Cl}$, a half-life of $(103\pm 7)\times {10}^3\mathrm{yr}$ was determined for ${}^{41}\mathrm{Ca}$. Adapted from [251].

Figure 44

Occurrence probability density function (OPDF) of solar energetic particles with the annual fluence $F>30\mathrm{MeV}$ exceeding the giving value. The triangles denote OPDF based on the data for the spacecraft. The circles correspond to the solar energetic particle (SEP) events derived from terrestrial cosmogenic and neutron-monitor data [modified after 516]. Open symbols indicate the measured or estimated values, whereas filled symbols indicate a conservative upper bound. Error bars bound the 68% confidence interval. The red hatched area encompasses the OPDF estimated in this work from ${}^{26}\mathrm{Al}$ in lunar samples. Adapted from [414].